Training-free, Perceptually Consistent Low-Resolution Previews with High-Resolution Image for Efficient Workflows of Diffusion Models

Wongi Jeong; Hoigi Seo; Se Young Chun

eess.IVApr 10, 2026

Training-free, Perceptually Consistent Low-Resolution Previews with High-Resolution Image for Efficient Workflows of Diffusion Models

Authors: Wongi Jeong, Hoigi Seo, Se Young Chun

Organizations: Dept. of Electrical and Computer Engineering, Seoul National University, Republic of Korea · INMC & IPAI, Seoul National University, Republic of Korea

Abstract

Image generative models have become indispensable tools to yield exquisite high-resolution (HR) images for everyone, ranging from general users to professional designers. However, a desired outcome often requires generating a large number of HR images with different prompts and seeds, resulting in high computational cost for both users and service providers. Generating low-resolution (LR) images first could alleviate computational burden, but it is not straightforward how to generate LR images that are perceptually consistent with their HR counterparts. Here, we consider the task of generating high-fidelity LR images, called Previews, that preserve perceptual similarity of their HR counterparts for an efficient workflow, allowing users to identify promising candidates before generating the final HR image. We propose the commutator-zero condition to ensure the LR-HR perceptual consistency for flow matching models, leading to the proposed training-free solution with downsampling matrix selection and commutator-zero guidance. Extensive experiments show that our method can generate LR images with up to 33% computation reduction while maintaining HR perceptual consistency. When combined with existing acceleration techniques, our method achieves up to 3 $\times$ speedup. Moreover, our formulation can be extended to image manipulations, such as warping and translation, demonstrating its generalizability.

Figures & tables

Figure 1: Motivation of Preview Generation. Most users of generative AI usually produce diverse candidate images through repeated trials from multiple seeds (or prompts) to obtain a desired result. Our goal is to accelerate this workflow by generating low-resolution images that share the same content as their high-resolution counterparts but can be produced much faster in a training-free manner.

Figure 2: Comparison between SR-upsampled and directly generated HR images. Using FLUX.1-dev, we (a) generated a low-resolution (LR, $256\times 256$ ) image followed by $4\times$ super-resolution (SR) to obtain a high-resolution (HR, $1024\times 1024$ ) image, and (b) directly generated an HR ( $1024\times 1024$ ) image. As shown in the close-up view, (a) fails to recover eye-region and fine fur details lost at the LR stage, whereas (b) preserves both global structure and fine details with high fidelity.

Figure 3: Overall framework. (Left, Top) Overview of our proposed framework. Sampling is first performed in the high-resolution (HR) space up to timestep $t_{D}$ , after which downsampling is applied using the selected downsample matrix from (Right, Top) . To maintain alignment with HR sampling, commutator-zero guidance is applied as shown in (Right, Bottom) . Finally, as illustrated in (Left, Bottom) , our method produces a low-resolution image with lower DreamSim score, which indicates better LR-HR consistency.

Base model	FLUX.1-dev	SD3.5-L
$\\|[\mathbf{D},v_{\theta}](x_{t},t)\\|\;(\pm 1\sigma)$	111.03 $(\pm 29.63)$	105.90 $(\pm 22.38)$

Table 1: Mean L2-norm of the commutator across models. We report the spatial-wise averaged L2-norm of the commutator over 100 samples for FLUX.1-dev and Stable Diffusion 3.5-Large (SD3.5-L). All models generate images or videos with 30 function evaluations (NFE). These results illustrate that flow matching models do not satisfy the commutator condition.

Method	Efficiency			Image quality	Perceptual sim.		Low-level sim.
	Latency(s) $\downarrow$	TFLOPs $\downarrow$	Speed $\uparrow$	PIQE $\downarrow$	DreamSim $\downarrow$	DiffSim $\uparrow$	PSNR (dB) $\uparrow$	FSIM $\uparrow$
FLUX.1-dev (30)	14.78	1800.49	1.00 $\times$	28.92	-	-	-	-
FLUX.1-dev (20)	10.01	1205.25	1.49 $\times$	34.78	7.25	0.8686	20.633	0.8268
Low-res.	4.94	602.96	2.99 $\times$	29.11	21.02	0.7352	11.936	0.6147
Naïve Down.	8.53	1029.43	1.75 $\times$	31.71	9.20	0.8584	18.221	0.7375
Ours	9.76	1178.30	1.53 $\times$	28.55	6.83	0.8721	21.182	0.7953

Table 2: Quantitative comparison on FLUX.1-dev and SD3.5-L. Using the FLUX.1-dev and Stable Diffusion 3.5-Large (SD3.5-L) models with

\text{NFE}=30

, we generate HR (

1024\times 1024

) reference images. We compare three variants: (i) reduced-NFE generation, (ii) LR (

512\times 512

) generation with the same NFE, and (iii) a naïve baseline applying nearest downsampling at timestep

t_{D}

. Each method is evaluated for computational efficiency, image quality, and consistency with the reference. Numbers in parentheses denote the NFE used; if omitted,

\text{NFE}=30

is applied as in the reference. We denote the best performance in bold and second-best performance with underline .

Method	Speed $\uparrow$	PIQE $\downarrow$	DreamSim $\downarrow$	PSNR (dB) $\uparrow$
FLUX.1-dev (30)	$1.00\times$	28.92	-	-
FLUX.1-dev (10)	$2.95\times$	32.82	16.97	16.427
Taylor only	$3.21\times$	28.10	9.17	18.667
Ours + Taylor	$3.05\times$	27.47	7.79	19.953

Table 3: Quantitative comparison on FLUX.1-dev with temporal-axis acceleration. We compare our approach combined with orthogonal temporal-axis acceleration. Our method achieves superior performance across all metrics, including image quality, perceptual similarity, and low-level similarity.

Figure 4: Qualitative comparison of our proposed method. While other simple alternatives often result in changes to composition, object size, or even color tone, our proposed approach synthesizes low-resolution images faster while preserving the composition and color fidelity of the original image. The prompts used for image generation are provided in the supplementary materials.

Figure 5: Generalization of commutator-zero guidance. We show that commutator-zero guidance can be expanded to other operations. For warping, a large kernel ( $128\times 128$ ) with correlation correction produces distortion, while our method effectively handles artifacts. For translation, naïve cause noticeable difference and unintended objects, whereas ours preserves image content.

	PIQE $\downarrow$	DreamSim $\downarrow$	PSNR $\uparrow$ (dB)
Nearest $\mathbf{D}$	29.72	10.39	18.069
Random $\mathbf{D}$	27.35	7.40	20.851
$\arg\max(\cdot)$ (Eq. 8 )	27.83	8.55	20.408
Full (Ours)	27.08	7.05	20.962

Table 4: Ablation study on

\mathbf{D}

selection. We compare three strategies for selecting

\mathbf{D}

: (i) nearest-neighbor downsampling, (ii) random sampling, and (iii)

\mathbf{D}^{*}

obtained via the

\arg\max(\cdot)

operation in Eq. 8 . The results show that our approach, which minimizes the commutator value, achieves the best performance.

CG	PIQE $\downarrow$	DreamSim $\downarrow$	PSNR $\uparrow$ (dB)
✗	31.89	8.56	19.115
✓	27.08	7.05	20.962

Table 5: Ablation study on commutator-zero guidance. Comparing results with and without commutator-zero guidance (CG) shows that while computational efficiency slightly decreases with correction, both perceptual similarity and PSNR increase, highlighting the effectiveness of our proposed method.

Panel of Table 6 — Table 6: Ablation study on hyperparameter $m$ and $\alpha$ . We investigate how varying $m$ and $\alpha$ affects performance. Increasing $m$ improves PSNR but also raises computational demand, showing a clear trade-off. Each $m$ exhibits an optimal step size $\alpha$ .

Figure 6: Changes in average commutator norm. We measure the change in commutator norm at timesteps

t_{D}

and

t_{D+m}

with and without applying commutator-zero guidance (CG). When CG is applied, the commutator norm decreases significantly, whereas without CG, it shows a statistically significant increase.

Figure 7: Cosine similarity analysis on velocity consistency of flow-matching models. We report the mean and standard deviation of cosine similarity between

v_{\theta}(x_{t_{D}},t_{D})

and

v_{\theta}(x_{t_{D+k}},t_{D+k})

over 500 samples. The similarity remains above 0.95 for

k\leq 5

, supporting the validity of the approximation.

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

	Prompt used in main qualitative results
1	pixelated Quantum entanglement Fentanyl digital banality, dishevelled, avant-garde multicultural diaspora humans contorted jumping twisted broken distorted lighting casts high contrast shadows lighting Hasselblad h6d - 400c, carl zeiss batis super close up 16mm f/ 2.8, ricoh r1
2	Small Dragon with Finnish elf characters, in a Finnish forest, in the style of korik kokiri, volumetric lighting, lively tableaus, mori kei, dark gold and green, vray, konica big mini. a figure and a mushroom stand among mushrooms in a forest, in the style of rendered in cinema4d, bill gekas, light gold and green, anime-inspired character designs.
3	Chinese man has invented a machine that can travel in time, High and short depth of field, stippling
4	shadoflectocyberchromos, You got a heart of stone, you can never feel.
5	Black Hole Sun + ultra high quality + beautiful colours + psychedelic + 3D rendered
6	A Stone Island X Cottweiler menswear collaboration editorial campaign shot by Johnny Dufort. Pa Salieu is modelling. Metallic fabrics. Mesh inserts. Two-tone materials. Chameleon fabrics. Color-flip materials. Technical treatments and washes to the materials. Night, Raining.

Appendix

Table S1 : The prompts used for the qualitative results in the main paper are ordered from the left column to the right column.

$k$	TFLOPs $\downarrow$	Speed $\uparrow$	PIQE $\downarrow$	DreamSim $\downarrow$	PSNR (dB) $\uparrow$
1	1178.30	1.53 $\times$	28.55	6.83	21.182
2	1277.56	1.41 $\times$	28.79	6.85	21.132
3	1376.84	1.31 $\times$	28.88	6.76	21.132

Appendix

Table S2 : Quantitative comparison across different values of

k

Figure S1 : Qualitative results on video diffusion. Using our preview generation framework with HunyuanVideo, we compare against an alternative approach that reduces NFE from 50 to 30. While Mammoth deviates from the full-step composition beyond a certain frame, our preview generation faithfully preserves the composition of the high-resolution content.

Figure S2 : Qualitative results of our proposed method at a lower resolution of $256\times 256$ .

Figure S3 : Qualitative comparison of our proposed method on FLUX.

Figure S4 : Qualitative comparison of our proposed method on FLUX.

Figure S5 : Qualitative comparison of our proposed method on SD3.5-Large.

Figure S6 : Qualitative comparison of our proposed method on SD3.5-Large.

Figure S7 : Qualitative comparison after integrating temporal-axis acceleration on FLUX.

Image Generation

Image generation studies models that create or edit images from text, examples, layouts, and other structured controls.

See all 463 papers

Aug 5, 2026

When does training on downscaled images yield the same gradients?

Diffusion transformers deliver strong image generation, but their training cost grows superlinearly with resolution. Recent work justifies training or sampling at reduced resolution on a spectral premise: at high noise, a downscaled latent preserves almost the full surviving signal. Whether a downscaled step also preserves the native training gradient signal, however, has remained unresolved. We reduce how that signal changes under downscaling to two terms: a noise-dependent term governed by the downscale ratio, which decays at high noise as the spectral premise predicts, and a σ-independent floor governed by the target grid's absolute token count, carried by the compute graph itself and removed by no noise level. The measured (route, σ) map corroborates the account and uncovers structure the spectral picture cannot express: on the 1024->768 route, a window (0.65 < σ< 0.95), predicted by no spectral criterion at any tolerance, where the downscaled gradient stays within a small margin of the native one. Training LoRA adapters with downscaled steps restricted to the routes and noise windows the map validates reduces training time by 14.6% at a fixed step budget while remaining near-native in weight space. Code is available at https://github.com/sorryhyun/anima_lora.

Seunghyun Ji

Independent researcher

Diffusion Transformer Diffusion Model Fine-Tuning Diffusion Model Acceleration Diffusion Models

Ask PDF

Jul 3, 2026

Perceptual Flow Matching for Few-Step Generative Modeling

We propose Perceptual Flow Matching (PFM), a simple yet effective framework for few-step generation in flow-matching models. Rather than performing velocity regression in the conventional VAE latent space, PFM supervises flow matching in a perceptual feature space using pretrained perceptual models. This simple change substantially improves the few-step generation capability of flow-matching models, reducing the number of sampling steps from 35-50 to 4-8 while preserving generation quality. Unlike existing acceleration and distillation approaches, PFM requires neither teacher models nor auxiliary score networks and can be integrated into standard flow-matching training pipelines with minimal modifications. Extensive experiments on image generation, video generation, and image editing tasks demonstrate that PFM consistently produces high-quality results while producing fewer artifacts than existing distillation-based methods. We further show that perceptual supervision shifts the regression minimizer from mean-seeking to mode-seeking, biasing predictions toward on-manifold modes that remain accurate under coarse few-step integration. Our results reveal that standard flow-matching training can naturally yield high-quality few-step generators when supervised in an appropriate representation space. We hope this insight inspires future research into representation-aware objectives for efficient generative modeling.

Chuyang Zhao, Yifei Song, Hongfa Wang +7

1Joy Future Academy · 2Fudan University · 3Tsinghua University +1

Flow Matching Perceptual Loss Image Generation Video Generation Flow-Based Generative Modeling

Ask PDF

Aug 7, 2026

HRDiT: Training-Free High-Resolution Image Generation with Off-the-Shelf Diffusion Transformer Models

Training-free text-to-high-resolution image generation has recently attracted growing research attention. However, existing studies on this task primarily focus on adapting off-the-shelf U-Net-based diffusion models to high resolutions, with limited progress on adapting off-the-shelf Diffusion Transformer (DiT) models despite their strong text-to-image generation capabilities at limited resolutions. In this work, we find two key challenges particularly hindering the application of off-the-shelf DiT models for high-resolution image synthesis in a training-free manner, namely, spatial disorder and long generation time. To address these challenges, we propose a novel method tailored to adapt off-the-shelf DiT models for high-resolution image synthesis. Extensive experiments show the efficacy of our method. Our code is available at: https://github.com/zylwithxy/HRDiT.

Yu Xue, Haoxuan Qu, Zhuoling Li +5

Lancaster University, Lancaster, UK · South China University of Technology, Guangzhou, China · NVIDIA AI Tech Centre, Singapore

Diffusion Model Adaptation Diffusion Transformer T2I Diffusion Models Diffusion Models

Ask PDF

Abstract

Image Generation

Explore similar work

When does training on downscaled images yield the same gradients?

Perceptual Flow Matching for Few-Step Generative Modeling

HRDiT: Training-Free High-Resolution Image Generation with Off-the-Shelf Diffusion Transformer Models